qa: Run all tests even if wallet is not compiled

[maflcko]

Currently the test_runner would exit if the wallet was not compiled into the Bitcoin Core executable. However, a lot of the tests run without the wallet just fine and there is no need to globally require the wallet to run the tests.

[maflcko]

Travis run for reference: https://travis-ci.org/bitcoin/bitcoin/jobs/426386820#L2937

[DrahtBot]

Note to reviewers: This pull request conflicts with the following ones:

• Drop support for getrawtransaction for confirmed tx without txindex #13931 (Drop support for getrawtransaction for confirmed tx without txindex by sipa)
• Lower default relay fees #13922 (Lower default relay fees by ajtowns)
• Bugfix: Only run bitcoin-tx tests when bitcoin-tx is enabled #12246 (Bugfix: Only run bitcoin-tx tests when bitcoin-tx is enabled by luke-jr)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

[lucash-dev]

This PR includes a commit that's also present in #14179 (fae497d). What's the relationship between the two PRs?

[maflcko]

This is based on #14179, since that is a requirement.

[Sjors]

Concept ACK. It's useful to (occasionally) run the test suite without wallet (and other stuff) compiled.

Quite a few tests are skipped this way, but that can be improved later. That might require pre-generated transactions (only when --disable-wallet), or the ability for a node to generate transactions without having full wallet functionality.

How will this interact with #10102 (multi process), which has one binary with and one without a wallet? cc @ryanofsky. I imagine a common test configuration involves one wallet binary to generate transactions as well as one node binary.

I tested on macOS (only the --disable-wallet variant).

WARNING! There is already a bitcoind process running on this system. Tests may fail unexpectedly due to resource contention!

wallet_hd.py skipped

Can you add one WARNING! at the top for each disabled feature, so it's more clear what caused the skipped lines?

[maflcko]

Quite a few tests are skipped this way, but that can be improved later.

Agree that this can be improved later. Skipping quite a few tests is better than skipping all tests.

[Empact]

nit: could invert the conditional / drop the continue

[laanwj]

Agree that this can be improved later. Skipping quite a few tests is better than skipping all tests.

👍

I do think this combination needs to be tested in Travis, though, or it's going to code-rot. At least I never run the tests without wallet.

[Sjors]

@laanwj the x86_64 Linux, No wallet Travis host already runs this:

[jnewbery]

Concept ACK

[laanwj]

utACK

[ryanofsky]

Note: Some code review comments on this PR can be found in #14179 instead of here, since this PR was originally based on #14179, but I think got accidentally merged first.